The present disclosure relates to semiconductor apparatuses and processes, and more particularly, to plasma mediated processes and plasma apparatuses suitable for ashing organic material from a substrate including a low k dielectric material.
Recently, much attention has been focused on developing low k dielectric thin films for use in the next generation of microelectronics. As integrated devices become smaller, the RC-delay time of signal propagation along interconnects becomes one of the dominant factors limiting overall chip speed. With the advent of copper technology, R has been pushed to its practical lowest limit so attention must be focused on reducing C. One way of accomplishing this task is to reduce the average dielectric constant (k) of the thin insulating films surrounding interconnects. The dielectric constant (k) of traditional silicon dioxide insulative materials is about 3.9. Lowering the dielectric constant (k) below 3.9 will provide a reduced capacitance and improve overall chip speed.
Low k dielectric materials used in advanced integrated circuits typically comprise organic polymers or oxides and have dielectric constants less than about 3.5. The low k dielectric materials can be spun onto the substrate as a solution or deposited by a chemical vapor deposition process. Important low k film properties include thickness and uniformity, dielectric constant, refractive index, adhesion, chemical resistance, thermal stability, pore size and distribution, coefficient of thermal expansion, glass transition temperature, film stress, and copper diffusion coefficient.
In fabricating integrated circuits on wafers, the wafers are generally subjected to many process steps before finished integrated circuits can be produced. Low k dielectric materials, especially carbon containing low k dielectric materials, can be sensitive to some of these process steps. For example, plasma used during an “ashing” step can strip both photoresist materials as well as remove a portion of the low-k dielectric film. Ashing generally refers to a plasma mediated stripping process by which residual photoresist and post etch residues are stripped or removed from a substrate upon exposure to the plasma. The ashing process generally occurs after an etching or implant process has been performed in which a photoresist material is used as a mask for etching a pattern into the underlying substrate or for selectively implanting ions into the exposed areas of the substrate. The remaining photoresist and any post etch or post implant residues on the wafer after the etch process or implant process is complete must be removed prior to further processing for numerous reasons generally known to those skilled in the art. The ashing step is typically followed by a wet chemical treatment to remove traces of the residue, which can cause further degradation of the low k dielectric, loss of material, and may also cause increase in the dielectric constant.
It is important to note that ashing processes significantly differ from etching processes. Although both processes may be plasma mediated, an etching process is markedly different in that the plasma chemistry is chosen to permanently transfer an image into the substrate by removing portions of the substrate surface through openings in a photoresist mask. The plasma generally includes high-energy ion bombardment at low temperatures and low pressures (of the order of millitorr) to remove portions of the substrate. Moreover, the portions of the substrate exposed to the ions are generally removed at a rate equal to or greater than the removal rate of the photoresist mask. In contrast, ashing processes generally refer to selectively removing the photoresist mask and any polymers or residues formed during etching. The ashing plasma chemistry is much less aggressive than etching chemistries and is generally chosen to remove the photoresist mask layer at a rate much greater than the removal rate of the underlying substrate. Moreover, most ashing processes heat the substrate to temperatures greater than 200° C. to increase the plasma reactivity, and are performed at relatively higher pressures (of the order of a torr). Thus, etching and ashing processes are directed to removal of significantly different materials and as such, require completely different plasma chemistries and processes. Successful ashing processes are not used to permanently transfer an image into the substrate. Rather, successful ashing processes are defined by the photoresist, polymer and residue removal rates without affecting or removing underlying layers, e.g., low k dielectric layers.
Studies have suggested that a significant contribution to low k dielectric degradation during photoresist removal processes results from the use of oxygen and/or nitrogen and/or fluorine containing gas sources for generating ashing plasmas. Although gas mixtures containing one or more of these sources efficiently ash photoresist from the substrate, the use of these gas sources has proven detrimental to substrates containing low k dielectrics. For example, oxygen-containing plasmas are known to raise the dielectric constant of low k dielectric underlayers during plasma processing. The increases in dielectric constant affects, among others, interconnect capacitance, which directly impacts device performance. Moreover, the use of oxygen-containing plasmas is generally less preferred for advanced device fabrication employing copper metal layers since copper metal is readily oxidized. Occasionally, the damage caused by oxygen containing plasmas is not readily detected during metrology inspection of the substrate after plasma processing. However, the damage can be readily demonstrated and observed by a subsequent wet cleaning process, as may be typically employed after plasma ashing, wherein portions of the carbon and/or hydrogen-containing low k dielectric material are removed. The removed portions of the dielectric material are a source of variation in the critical dimension (CD) of the feature that is frequently unacceptable and impacts overall device yield. Moreover, even if a wet clean process is not included, the electrical and mechanical properties of the dielectric material may be changed by exposure to the oxygen-free plasmas thereby affecting operating performance. It is believed that carbon is depleted from the dielectric material during the oxygen containing plasma exposure.
Ideally, the ashing plasma should not affect the underlying low k dielectric layers and preferentially removes only the photoresist material. The use of traditional dielectrics (higher k values) such as SiO2 provided high selectivity with these gas sources and was suitable for earlier device generation. However, in order to minimize damage to the low k dielectric, oxygen free and nitrogen free plasma processes have been developed. One such process includes generating plasma from a gas mixture comprising helium and hydrogen resulting in a different mechanism of removal. The oxygen free and nitrogen free plasma such as the plasma formed from helium and hydrogen is less aggressive and does not completely react with the photoresist in the traditional sense. Rather, it is believed that the plasma renders portions of the photoresist to be removable such as by sublimation. As a result of the mechanism of removal, while effective for removing photoresist material from the substrate, the plasma exposure tends to deposit large bodies of the sublimed or removed photoresist and byproducts within the processing chamber and in areas downstream from the plasma process chamber such as in the exhaust lines and any components therein. The buildup of ashing material can lead to short mean-time-between-clean (MTBC) and frequent rebuild/replacement of vacuum hardware resulting in loss of throughput and increased costs of ownership. Additionally, deposits of photoresist material within the process chamber that are located above the plane of the substrate can lead to particulate contamination on the substrate, thereby further affecting device yields.
An additional problem with oxygen free and nitrogen free plasmas is the non-uniformity of the plasma exposure. Since these plasmas are less aggressive, non-uniformity is a significant issue. Some downstream plasma ashers have a narrow diameter orifice plasma tube in which the plasma is generated. The diameter of the substrate is generally much larger than the diameter of the plasma tube orifice. As such, baffle plates are typically positioned near the plasma tube outlet to deflect the plasma as it enters the process chamber such that the species in the plasma are uniformly dispersed across the substrate. However, it has been found that the less aggressive plasmas exhibit less ashing efficiency as reactive species are dispersed from the center point of the baffle plate to its outer edge resulting in hot spots, i.e., areas of non-uniformity. For example, it has been discovered that hydrogen radicals generated within a plasma decrease in ashing efficiency as the hydrogen radicals travel from the center most impingement point on the baffle plate in the axial flow reactor to the outer edges of the baffle plate. In chamber designs where the diameter of the wafer is comparable to that of the plasma tube, non-uniformity of radicals can be mitigated in other ways.
Another problem with oxygen free and nitrogen free plasmas concerns endpoint detection. Traditional endpoint detection methods and apparatus are not suitable for use with these types of plasmas. For example, as in the case of plasma formed from a hydrogen and helium gas mixture, no optically excited species are emitted that are sufficient to generate a signal suitable for endpoint detection.
Accordingly, there remains a need for improved processes and apparatuses for generating oxygen and nitrogen free plasmas for use with low k dielectrics.